Concern with potential toxicity due to the widespread use of unlicensed and off label drugs in children has led to regulatory changes aimed to strengthen the evidence base for paediatric drugs. This thesis examines paediatric randomised controlled trials (RCTs), the highest level of evidence, and assesses them in relation to global child health. A systematic review was performed using validated methods to search three major databases for paediatric RCTs published in 2007. More than 600 RCTs were identified involving more than 100,000 children. The RCTs appear to study the appropriate clinical areas however few studies involved neonates. The RCTs also seem to be of good methodological quality with a mean Jadad score of 3.22. The reporting of RCTs that involve both adults and children needs to be improved to add to the evidence base of paediatric medicines. More attention is also needed on the reporting of safety information from the RCTs to provide useful toxicity data. Although severe and moderate ADRs were seen in 25% of the RCTs, few RCTs (12%) established safety monitoring committees (SMCs). SMCs are vital to ensure patients in paediatric RCTs are protected from toxicity. The burden of childhood disease is heaviest in low and middle income countries (LMIC). A minority of the RCTs were performed in LMIC, although they are increasingly globalised. RCTs conducted in LMIC appear to have lower methodological quality, and reported less well on ethical approval and adverse events. In conclusion high quality, ethical paediatric RCTs should add to the evidence base for paediatric medicines. However they should correspond with the health needs of children on a global basis.

Background: There have been many epidemiological studies into the influence of exposure to the most toxic elements on the risk of cancer in the workplace, mainly due to the exposure of certain occupational groups, or perhaps in populations near industrial sources. Toxic elements include arsenic, copper, nickel, and uranium; and many more of these elements have been shown to increase the risk of several different types of cancers in these highly-exposed groups. Many of these elements naturally exist in the soil, and the health impact of these levels of environmental exposures on the general population has received little attention to date possibly due to the belief that soil concentrations of these elements are too low to cause harm to the general population. Therefore, the long-term effect of such chronic exposure to metals in the soil remains unclear. Aims and objectives: The goals are to utilise a new resource known as THIN-GBASE for conducting a series of environmental epidemiological studies to test the hypothesis that BCC, lung and GIT cancers are associated with high exposure to certain low-level metals in soil. We sought to use this resource in determining which soil metals should be tested for predicting each of the cancer outcomes. Methods: For BCC, an ecological study was initially undertaken to assess the overall regional variation in BCC to provide national and contemporary breakdowns of incidence rates across the UK. The primary exposure of interest for BCC was low-level soil arsenic, and we therefore quantified soil arsenic exposure levels based on the UK national safety limits for arsenic [i.e. As-C4SLs = 35 mg/kg]. A population-based cohort study was conducted to quantify the risks associated between the development of BCC and increasing levels of exposure to soil arsenic. For lung cancer, a two-stage process was adopted: 1) data mining analysis using the correlation-based filter selection model was used to find the restricted set of soil metals were best predictors for lung cancer; and 2) a prospective cohort study was use where these sets of elements were fitted together (adjusted for confounding variables) in a multivariable Cox proportional-hazards model to determine the risks associated between the development of lung cancer, with increasing levels of exposure to each specific element. For GIT cancers, a three-stage process was adopted: stages 1 and 2 used a similar methodology for the lung cancer study. In stage 3, all GIT cancers were divided into three broader outcomes i.e. upper GIT (includes mouth & oesophagus), stomach (as standalone) and colorectal (includes small, large, rectum and anal canal) cancers. A multivariate competing risk survival model was adjusted for the three different GIT cancers as competing events to identify associations between any of the selected group of metals found in stage 1 and GIT-specific cancers. Results: For BCC, the findings for the ecological study show that overall EASRs & WASRs for BCC in the UK was 98.6 and 66.9 per 100,000 person-years, respectively. It indicates a large geographical variation in age-sex standardised incidence of BCC with the South East having the highest incidence of BCC (202.7/100,000 person-years), followed by South Central (193.5/100,000 person-years) and Wales (185.7/100,000 person-years). Incidence rates of BCC were substantially higher in the least socioeconomically deprived groups. It was observed that increasing levels of deprivation led to a decreased rate of BCC (p < 0.001). In terms of age groups, the largest annual increase was observed among those aged 30-49 years. Assessment for soil arsenic indicated that individuals living in areas with concentrations ≥35mg/kg significantly had an increased hazard of developing BCC (35-70mg/kg: adjusted HR 1.08, 95% CI: 1.02-1.14; ≥70mg/kg: adjusted HR 1.17, 95% CI: 1.09-1.28). Urban residents with the highest exposure of soil arsenic had the greatest risk of developing BCC (≥ 70.0 mg/kg: HR 1.18, 95% CI: 1.06-1.36). For lung cancer, the correlation-based filter selection model identified aluminium, lead and uranium as the appropriate set of exposures for modelling lung cancer risk. Complete adjustments of hazards model showed evidence of an increased risk of developing lung cancer with elevated concentrations for only soil aluminium at medium levels ranging between 47,000-61,600mg/kg. Urban residents with the highest exposure of soil aluminium had the greatest risk of developing lung cancer (≥ 61,600mg/kg: HR 1.12, 95% CI: 1.04-1.22). For GIT cancers, the correlation-based filter selection model identified seven elements i.e. aluminium, phosphorus, zinc, uranium, calcium, manganese, and lead, as the appropriate set of exposures for predicting GIT cancer risk. The complete adjustment for hazards model indicated that the risk of developing overall GIT cancers were significantly associated with elevated exposure levels of soil phosphorus only (873-1,127mg/kg: HR 1.08, 95% CI: 1.02-1.14; 1,127-1,456mg/kg: HR 1.07, 95% CI: 1.01-1.13; and ≥1,145mg/kg: HR 1.07, 95% CI: 1.01-1.13). There were no consistent relationships identified between any of the selected groups of elements and the GIT-specific cancer outcomes when adjusting for different GIT cancers as competing events. Conclusion: There appears to be slight evidence of BCC, respiratory and GIT cancer risk with elevated exposure to soil arsenic, aluminium and phosphorus, respectively. The series of investigations conducted for this research are one of the first, if not, contemporary UK-based study to present novel estimates for a group of ill-defined pollutants. This research demonstrates that linking geochemical data with electronic primary care medical records can be a valuable approach of proving whether long term exposure to low-level soil contaminants may have a health consequence in the population.

Multimodal analgesia has been proposed as a useful strategy to reduce postoperative pain while decreasing opioid consumption and thus opioid adverse events. Gabapentin is one such agent although previous results have been heterogeneous. This thesis aimed to review randomised controlled trials of gabapentin for reducing pain, opioid adverse effects and the haemodynamic response to intubation while attempted to predict clinical effectiveness from these trials using meta-regression. Extending this principle, we evaluated other multimodal analgesic agents to identify whether heterogeneity could be explained by various clinical and methodological covariates. Our gabapentin review included 133 randomised controlled trials and demonstrated its efficacy in reducing pain scores, opioid consumption and opioid adverse events such as nausea, vomiting and pruritus. However, gabapentin increased the risk of sedation. Gabapentin was effective at reducing the haemodynamic response to intubation in 29 randomised controlled trials although trials failed to report on clinically relevant outcomes. Gabapentin exhibited no pre-emptive analgesic effect in 4 randomised controlled trials. There was evidence of considerable statistical heterogeneity on meta-analysis of gabapentin for pain scores and 24-hour morphine consumption. Meta-regression analysis showed however that baseline risk predicted the majority of the heterogeneity between studies. Extending this approach to other multimodal analgesics from 344 randomised controlled trials; we demonstrated this was true for analgesic agents in general. In addition to baseline risk, methodological limitations, especially inadequate allocation concealment, explained some of the residual heterogeneity. There was evidence of funnel plot asymmetry for most analgesic agents, suggesting publication bias. However, this may be a product of trials with higher baseline risk having larger standard errors, rather than true publication bias. Indeed, when we simulated meta-analyses with no publication bias, with both effect size and standard deviations dependent on baseline risk, funnel plot asymmetry was still evident (p < 0.001). Therefore, conventional funnel plots may be an unsuitable method of detecting publication bias where baseline risk predicts between-study heterogeneity. We present an alternative method using meta-regression residuals that corrects funnel plot asymmetry in the presence of no publication bias. Finally, due to concerns that methodological limitations exaggerated effect estimates, we used trial sequential analysis to determine whether sufficient low risk of bias evidence exists to reject type I and type II errors in the analyses of analgesic adjuncts. We demonstrated there is currently insufficient evidence from low risk of bias trials to be confident of the efficacy of the majority of analgesic adjuncts.

The analgesic effects of cannabis-based medicines are widely described and elevation of the endocannabinoids (ECLs), by inhibition of endocannabinoid metabolism, also produces analgesia. Measurement of endocannabinoids (ECLs) and metabolites, in low weight biological tissue, provides an important tool for investigating the potential therapeutic effects of this receptor system. The main aims of this thesis were to develop an analytical method to measure ECLs, and cyclooxgenase-2 (COX-2) metabolites of anandamide (AEA) and 2-arachidonoylglycerol (2-AG), and to apply the method to determine their role in nociceptive processing in models of acute and chronic inflammatory pain. A liquid chromatography-tandem mass spectrometry (LC-MS/MS) method, which allowed for the simultaneous measurement of AEA, oleoylethanolamide (OEA), palmitoylethanolamine (PEA) and 2-AG, as well as COX-2 metabolites of AEA and 2-AG was developed and validated. This method was demonstrated to be able to quantify ECLs and related compounds in rat hindpaw, brain, spinal cord and cell samples. The effects of TRPV1 activation on levels of intracellular calcium [Ca2+] and ECLs in sensory nerves were investigated. The TRPV1 agonist, capsaicin, dose dependently increased [Ca2+], which was blocked by the TRPV1 antagonist iodoresiniferatoxin (IRTX). Although synthesis of ECLs is activity and calcium dependent, there was no relationship between the TRPV1-mediated increases in [Ca2+], and levels of intracellular or extracellular ECLs. The mechanism of sensitization of primary afferent nociceptors by the proinflammatory cytokine tumor necrosis factor-α (TNFα), which is heavily implicated in inflammatory responses, was also studied. TNFα facilitated capsaicin-evoked calcium responses, an effect which was mediated by p38MAPK-induced phosphorylation. The ability of inhibition of fatty acid amide hydrolase (FAAH) to modulate inflammatory pain responses, and the sites at which ECLs were elevated, was studied in the carrageenan model of inflammatory pain. This model was considered ideal for the investigation of the role of alternative metabolism of ECLs by COX-2 as inflammatory pain is associated with an up-regulation of COX-2. The effects of acute and repeated administration of URB597 (0.3 mgkg-1), a selective inhibitor of FAAH metabolism of ECLs, on carrageenan-evoked changes in weight bearing and hindpaw oedema were determined. Behavioural effects were considered in parallel with the changes in levels of ECLs and COX-2 metabolites of AEA and 2-AG in the hindpaw and spinal cord. Acute administration of URB597 delayed the onset and attenuated carrageenan-induced hyperalgesia, an effect which was associated with increased levels of AEA, OEA and PEA in the spinal cord. By contrast, carrageenan-induced hyperalgesia was not altered by repeated administration of URB597 and levels of AEA in the spinal cord were not significantly increased by this treatment. There was no evidence for the metabolism of ECLs via COX-2 in the hindpaw, or spinal cord, of carrageenan-treated rats, under basal conditions or following inhibition of FAAH. The role of the endocannabinoid system in a model of osteoarthritis (OA) pain was also investigated. Intra-articluar injection of the chondrocyte inhibitor monosodium iodoacetate (MIA, 1 mg/50µl) produced histological changes reminiscent of joint damage in OA patients, and a time dependant decrease in paw withdrawal thresholds and weight bearing on the injured (ipsilateral) side. These behavioural pain responses were accompanied with an increase in ECLs in the spinal cord, suggestive of a role of ECLs in modulation of OA-induced pain.

Introduction and aim It has been suggested that the employment of pharmacists in general practice might moderate the growth in prescribing costs. However, empirical evidence for this proposition has been lacking. The aim of this study was to evaluate a controlled trial of pharmacist-led intervention in general practice to determine whether intervention practices made savings relative to controls and if so, exactly how these savings were made and whether quality of prescribing was maintained. Since this process of rationalisation has implications for patients, an additional aim was to explore the views of patients on changes made to their medication. Methods The study was an evaluation of an initiative set up by Doncaster Health Authority. Eight practices received intensive input from five pharmacists for one year (September 1996 to August 1997) at a cost of £163 000. Changes in prescribing patterns were investigated using Prescribing Analysis and CosT (PACT) data by comparing these practices with eight individually matched controls for both the year of the intervention and the previous year. A postal survey of 314 patients who had undergone a change in medication between October 1997 and January 1998 was used to explore patient views. Results The evaluation showed that the rise in prescribing costs for intervention practices was significantly lower than for control practices (p=0.02S). Had the cost growth of the intervention group been as high as that of the controls, their total prescribing expenditure would have been around £347 000 higher. Detailed analysis showed that these savings were achieved by controlling both prescribing volume and cost per unit volume in areas believed to be without detriment to patient care. The majority of patients were reasonably satisfied or very satisfied with the way in which they found out about their medication change and satisfaction was positively associated with being told why the change was taking place, being given a choice and being told by the GP, a practice pharmacist or by letter. Conclusions Compared with previous studies, this evaluation has advantages in the fact that a control group was used to compare changes in prescribing patterns. The evaluation has shown that the use of pharmacists controlled prescribing expenditure sufficiently to off-set the costs of their employment. Results of the patient survey indicated that patients were not so much concerned about changes in medication per se, but rather the manner in which it was conveyed to them. These results have important implications for the control of prescribing costs in primary care. However, this study took place in motivated practices that had relatively high prescribing costs and this may limit the generalisability of the results.

This thesis attempts to determine the role of 5-hydroxytryptamine (5-HT) in rodent models of anxiety. Using the elevated X-maze it was possible to detect dose-dependent anxiolytic effects of diazepam and dose-dependent anxiogenic effects of FG 7142 (a β-carboline derivative) and idazoxan. The fear-potentiated acoustic startle paradigm also detected the effects of these compounds but the results were neither dose-dependent nor very reproducible. Using the X-maze the 5-HT1A receptor partial agonist ipsapirone had no effect after either acute or chronic treatment, while both the 5-HT2 receptor antagonist, ritanserin, and the 5-HT3 receptor antagonist, ondansetron, had anxiolytic activity after chronic treatment. Using in vivo mirodialysis it was demonstrated that extracellular levels of 5-HT in the ventral hippocampus increase when the animal is on the X-maze, and both diazepam and F 2692 (1-(3'-trifluoro-methylphenyl) 1.4- dihydro 3-amino 4-oxo 6-methyl pyrldazine) reduced the increased 5-HT levels and produced an anxiolytic behavioural profile in the same animal. Thus the inhibition of increased 5-HT release may be important for anxiolytic activity. However, ipsapirone also reduced the increased 5-HT but did not produce an anxiolytic profile. The lack of an anxiolytic effect may be the result of postsynaptic 5-HT1A receptor stimulation. Rats reared in isolation immediately post-weaning (21 days of age) displayed enhanced locomotor activity and an anxiogenic profile on the X-maze. This anxiogenic profile was neither reversed by resocialisation of the isolation-reared rats nor produced by isolation of adult socially housed animals, indicating a permanent developmental change. Isolation-reared rats had reduced stimulated release of 5-HT measured in the frontal cortex, indicating reduced presynaptic function and enhanced responsiveness to agonists acting at postsynaptic 5-HT1A and 5-HT2 receptors, suggesting supersensitivity at these sites. Overall, 5-HT appears to be involved in anxiety with 'anxious' behaviour either on the X-maze or by isolation-rearing causing changes in 5-HT function. Thus one of the mechanisms of action of established and putative anxiolytic and anxiogenic compounds may be via modulation of the ascending dorsal raphe serotonergic neuronal pathways. All work in this thesis was undertaken under Home Office Personal Licence PIL 70/02103 and Project Licences PPL 40/0380 (Nottingham University) and PPL 70/00349 (SmithKline Beecham).

Mechanisms responsible for resistance to β-Iactam antibiotics were investigated in clinical isolates of Bacteroides spp., the most common anaerobic Gram negative pathogen. Among 108 isolates of Bacteroides spp. obtained from clinical material in Nottingham, 69 (64%) were identified as Bacteroides fragilis. Approximately one-fifth of the Bacteroides spp. produced elevated levels of β-Iactamases, and many of these strains showed increased resistance to β-Iactam antibiotics usually regarded as β-Iactamase stable. Four β-Iactamase types were identified: Type one was represented by zinc dependent metallo-β-lactamases that hydrolysed cefoxitin, latamoxef and imipenem. Type two displayed intermediate to high specific activity in tests with nitrocefin as substrate and hydrolysed cefoxitin and latamoxef, but not imipenem. Type three exhibited intermediate levels of specific activity and caused reduced susceptibility to cefoxitin, latamoxef and imipenem, although they hydrolysed these antibiotics inefficiently. Type four probably represented enhanced production of the β-Iactamases characteristic of most bacteroides strains. Organisms producing this enzyme normally remained susceptible to cefoxitin, latamoxef and imipenem, and the enzyme was fully susceptible to β-Iactamase inhibitors, including clavulanic acid. Several strains were detected that exhibited reduced susceptibility to imipenem that was not related to metallo-β-Iactamases production. In contrast, one strain that produced a metallo-β-lactamase remained fully sensitive to imipenem as judged by conventional titration. Investigation of isolates producing metallo-β-lactamase and several similar ones encountered in an earlier study showed a correlation between the degree of resistance to imipenem and specific imipenemase activity. In an attempt to elucidate reduced susceptibilty to imipenem that was not related to metallo-β-lactamase production, cell envelope properties and penicillin-binding proteins (PBPs) of selected strains were investigated. Studies of outer membrane proteins and lipopolysaccharide composition of B.fragilis strains unexpectedly showed that metallo-β-Iactamase producers displayed unusual cell envelope profiles. Those strains that showed reduced susceptibility that was not associated with β-Iactamase appeared to be normal, although two of these strains displayed abnormally high crypticity values. Between three and six PBPs were visualised in tests with 3H-benzylpenicillin. PBPs 1 to 3 were present in all strains, but were observed in imipenem-resistant strains only when tests were carried out in the presence of β-lactamase inhibitors. Competitive binding experiments indicated that imipenem showed affinity for the high molecular weight PBPs of sensitive B.fragilis. The low molecular weight PBPs 4 to 6 were detected intermittently and only in certain strains, notably those that exhibited reduced susceptibility to imipenem. PBP 6 was found only in strains showing non-enzymic resistance. The results suggest that several different types of resistance to β-lactam antibiotics are circulating in clinical isolates of Bacteroides spp. in Nottingham: firstly, strains that produce a metallo-β-lactamase that hydrolyses 'β-lactamase-stable' compounds, including imipenem; secondly, strains producing enzymes that hydrolyse cefoxitin and latamoxef, but not imipenem; thirdly, strains that possess a permeability barrier that affects imipenem as well as other β-Iactam antibiotics; fourthly, strains with altered penicillin-binding proteins; and fifthly strains that produce raised amounts of 'normal' β-lactamase. The latter strains slowly hydrolyse imlpenem and other β-lactam compounds and this may be a factor in the reduced susceptibility to these agents. Clinical isolates that possess these resistance mechanisms may appear susceptible according to conventional break-point criteria in vitro. Never the less, they exhibit considerably reduced susceptibility to β-Iactam agents and the therapeutic implications of this need investigating. At present such strains represent a relatively small proportion of clinical isolates of Bacteroides spp., but the prevalence of resistance needs to be carefully monitored.

Inter-individual variation in pharmacokinetic parameters of drugs can have profound effects on drug safety in children. Midazolam and morphine are among the most commonly used drugs in critically children. Theophylline has seen several cycles of enthusiasm and unpopularity over the years, although oral theophylline is now rarely used, IV aminophylline is still used regularly in severe asthma. These drugs are metabolised by hepatic enzymes (CYP3A4, CYP1A2 and glucuronidation) which have variable expression. Three systematic reviews were conducted in order to explore the inter-individual variation of clearance of these drugs in children. The first systematic review evaluated the inter-individual variability of midazolam clearance in children. Midazolam is predominantly metabolised by CYP3A4. Twenty two PK studies were identified. The mean clearance of midazolam varied between 0.78 to 3.5 ml/min/kg in neonates and 1.1 to 15 ml/min/kg in children. Age was a statistically significant predictor of clearance (p < 0.05). Critical illness was however not a statistically significant predictor of midazolam clearance after adjusting for other covariates (p=0.279). There was a statistically significant difference between the coefficient of variation of midazolam clearance in preterm neonates (91%) and children (40%) (p=0.002). However, there was no significant difference between the CV in critically ill and non-critically ill children. A second systematic review evaluated the variability of theophylline clearance. Theophylline is metabolised by CYP1A2. Twenty nine studies were identified. Mean clearance of theophylline varied between 0.2 and 2 ml/min/kg. Age was a significant predictor of theophylline clearance (p<0.05). There was, however, no significant difference between the CV of theophylline in any age group. The CV of theophylline clearance was not significantly different between critically ill (35%) and non-critically ill (39%) (p=0.403). A sub-analysis of children also did not show any significant difference between critically ill and non-critically ill (p=0.418). A third systematic review evaluated the variability of morphine clearance in children. Morphine is metabolised by UGT. Twenty studies were identified. The mean clearance of the studies identified varied between 2 and 16 ml/min/kg in neonates and 19 to 52 ml/min/kg in children. Critical illness was not a statistically significant predictor of morphine clearance. Analyses of the limited data showed no statistically significant differences in CV between any age groups. There was also no statistically significant difference between the CV in critically ill and non-critically ill children. In all the studies, a major limitation was the limited number of PK studies in children. Invasive studies should be avoided in children therefore, a final systematic review evaluated the invasiveness of PK studies over two decades. The number of blood samples collected per child was significantly lower in studies carried out between 2004-2014 than those between 1981-1990 (p=0.013). Furthermore, the total volume of blood collected in 24 hours for PK studies was significantly lower in new decade than old (p=0.025). However, there was no difference in the volume of blood collected per sample. There were 35 population PK studies, all of which were new studies. The median number of blood samples in population PK studies (median 6, [IQR: 4-9]) was significantly lower than non-population PK studies (median: 8, [IQR: 6-10]) (p=0.007). In conclusion, age is a risk factor for inter-individual variation of midazolam clearance in children. It is also an important predictor of midazolam, morphine and theophylline clearance in children. Therefore, age appropriate dosing is important. More PK studies are required to determine the effect of critical illness on the variability of clearance of these drugs. The utilisation of population PK methods should be encouraged to minimise invasiveness of PK studies. New methodologies for reducing sample volumes and frequency should be considered in all studies.

Several of the available antiepileptic drugs (AEDs) were approved in the last 25 years. These new generation AEDs have not been shown to be more effective than the old ones and their safety profile have not been explored sufficiently in pharmacovigilance studies. As reported in chapter 2, prospective cohort studies are the most common pharmacovigilance study methods, with adverse drug reactions (ADRs) often elicited with questionnaires or checklists. A systematic review to identify all published AED side effects rating scales, reported in chapter 3, identified nine AED ADR rating scales. Two of these, The Hague Side Effect Scale (HASES) and the Paediatric Side Effect Questionnaire (PESQ), are paediatric specific. A systematic review of AED utilisation rate reported in chapter 4, shows an increasing utilisation of levetiracetam and lamotrigine reported as the most frequently utilised new generation AED in paediatrics. Systematic reviews of the safety of both drugs in children, reported in chapters 5 and 6, identified rash (7.3%) and behavioural problems (10.9%) as the most common ADRs associated with lamotrigine and levetiracetam respectively. They were also the most common reasons for the discontinuation of treatment. In chapter 7, Stevens-Johnson Syndrome (SJS) and Toxic Epidermal Necrolysis (TEN) risks are reported to be significantly higher when lamotrigine was co-prescribed with sodium valproate (43%) than when either carbamazepine (8%) or phenobarbital (8%) were co-prescribed with sodium valproate. SJS/TEN also occurred more frequently with sodium valproate and lamotrigine co-medication than other non-cutaneous ADRs (19%). Being the more recent paediatric AED ADR rating scale, the PESQ was selected for the elicitation of ADRs in a prospective cohort study of AED safety in children reported in chapter 8. Half of the 124 participants in the study received levetiracetam, either as monotherapy or polytherapy. There were significantly fewer ADRs with levetiracetam than either carbamazepine or sodium valproate monotherapy. The risks of drowsiness, fatigue and weight gain were significantly higher with levetiracetam polytherapy than monotherapy (p < 0.05). Attention difficulties, aggression and decreased concentration were significantly lower with valproate polytherapy (p < 0.05). The common ADRs associated with AEDs are discussed in chapter 9. In conclusion, lamotrigine and levetiracetam are increasingly being used for the treatment of epilepsy in children. Lamotrigine may cause severe rash, especially when co-administered with valproate; while levetiracetam is a common cause of behavioural problems. In order to compare the safety profile of AEDs adequately, large multicentre paediatric safety studies are required.

Adverse drug reaction (ADR) reports of countries varies due to differences in the prevalence of diseases and hence the types of drugs used. ADRs are a major health and economic burden worldwide. National health authorities monitor the safety of medicines to protect consumers from the hazards of drugs. ADR databases are also maintained from where reports are regularly evaluated to detect signals of new ADRs and determine the increase of those already known. A review of paediatric and general population studies conducted on ADRs from national ADR databases was carried out. The majority of studies identified were from countries in Europe and North America, and only one study on the general population was conducted from the Ethiopian ADR database in Africa. No paediatric study was identified in Africa. Skin reactions associated with antiretroviral drugs were the most frequent ADRs in the study conducted from the Ethiopian ADR database. Anti-infective agents, mostly vaccines, were mostly associated with the ADRs in children in Europe and Latin America, and drugs used for treating attention deficit hyperactivity disorders (ADHD) were implicated with the ADRs reported for children in North America. The ADR databases of Ghana and Nigeria were analysed to evaluate the ADRs reported for children and adults. The fatalities reported and the associated drugs in the two databases were also evaluated. The ADR reporting rates for children and the general population in Ghana and Nigeria were lower than the corresponding rates observed in the review. The majority of the ADRs in Nigerian adults were reported for antiretroviral drugs, and most of those who died suffered Stevens Johnson syndrome with antimalarials as the suspect drugs. ADRs reported for Nigerian children were mainly skin reactions associated with antibiotics. Most of the reported fatalities resulted from renal failure, linked with suspected contaminated teething mixtures. Antimalarials and anthelmintics were mostly associated with the ADRs in Ghanaian adults. Most of the reported fatalities resulted from Stevens Johnson syndrome. ADRs in Ghanaian children were mostly associated with vaccines. The majority of the reported deaths resulted from unknown causes linked with antimalarials.